| Characteristic | Overall1 | Acute Hypoxaemic Respiratory Failure | |
|---|---|---|---|
| No (PF >300)1 | Yes (PF <300)1 | ||
| Number of Patients | 1,560,221 | 734,115 | 826,106 |
| Median Age in Years (IQR) | 66 (53-75) | 65 (49-75) | 67 (55-76) |
| Age Category, Years | |||
| <44 | 241,138 (15%) | 141,470 (19%) | 99,668 (12%) |
| >84 | 112,518 (7.2%) | 54,908 (7.5%) | 57,610 (7.0%) |
| 45-64 | 474,408 (30%) | 217,225 (30%) | 257,183 (31%) |
| 65-84 | 729,623 (47%) | 319,327 (44%) | 410,296 (50%) |
| Gender | |||
| Female | 641,352 (41%) | 326,027 (44%) | 315,325 (38%) |
| Male | 918,260 (59%) | 407,783 (56%) | 510,477 (62%) |
| Intersex/Indeterminate | 390 (<0.1%) | 205 (<0.1%) | 185 (<0.1%) |
| Unknown | 219 (<0.1%) | 100 (<0.1%) | 119 (<0.1%) |
| Median APACHE II Score (IQR) | 15 (11-21) | 14 (10-18) | 17 (13-23) |
| Median APACHE III Score (IQR) | 52 (39-68) | 47 (35-61) | 56 (43-74) |
| Median ANZROD (IQR) | 0.09 (0.18) | 0.06 (0.13) | 0.12 (0.21) |
| Median SOFA (IQR) | 4 (2-6) | 3 (1-4) | 5 (3-7) |
| Admission Diagnosis | |||
| Medical | 478,338 (31%) | 176,685 (25%) | 301,653 (37%) |
| Post-Operative | 432,367 (28%) | 243,005 (34%) | 189,362 (23%) |
| Sepsis | 113,516 (7.5%) | 47,693 (6.7%) | 65,823 (8.2%) |
| Trauma/Neurosurgery | 164,707 (11%) | 105,030 (15%) | 59,677 (7.4%) |
| Cardiac Surgery | 330,028 (22%) | 139,524 (20%) | 190,504 (24%) |
| COVID Penumonitis (Proven) | 4,451 (0.3%) | 249 (<0.1%) | 4,202 (0.5%) |
| Admission Source | |||
| Emergency Department | 379,997 (24%) | 159,769 (22%) | 220,228 (27%) |
| Operating Theatre/Recovery | 897,931 (58%) | 474,998 (65%) | 422,933 (51%) |
| Ward | 188,779 (12%) | 61,825 (8.4%) | 126,954 (15%) |
| ICU, Same Hospital | 2,088 (0.1%) | 769 (0.1%) | 1,319 (0.2%) |
| Other Hospital | 90,715 (5.8%) | 36,435 (5.0%) | 54,280 (6.6%) |
| Direct from Home | 711 (<0.1%) | 319 (<0.1%) | 392 (<0.1%) |
| Hospital Type | |||
| Tertiary | 738,248 (47%) | 339,133 (46%) | 399,115 (48%) |
| Metropolitan | 243,919 (16%) | 99,911 (14%) | 144,008 (17%) |
| Rural / Regional | 145,760 (9.3%) | 58,204 (7.9%) | 87,556 (11%) |
| Private | 432,294 (28%) | 236,867 (32%) | 195,427 (24%) |
| Invasive MV | 294,060 (46%) | 113,665 (37%) | 180,395 (54%) |
| IMV Day 1 | |||
| 1 | 272,472 (17%) | 105,681 (14%) | 166,791 (20%) |
| 2 | 444,137 (28%) | 246,042 (34%) | 198,095 (24%) |
| NULL | 843,612 (54%) | 382,392 (52%) | 461,220 (56%) |
| NIV | 74,382 (13%) | 13,723 (4.9%) | 60,659 (20%) |
| ECMO | 1,630 (0.3%) | 257 (<0.1%) | 1,373 (0.5%) |
| Tracheostomy | |||
| 1 | 8,962 (0.6%) | 3,072 (0.4%) | 5,890 (0.7%) |
| 2 | 553,685 (35%) | 274,161 (37%) | 279,524 (34%) |
| NULL | 997,574 (64%) | 456,882 (62%) | 540,692 (65%) |
| 1 n; Median (25%-75%); n (%); Mean (SD) | |||
Epidemiology and Clinical Outcomes of Patients with Acute Hypoxaemic Respiratory Failure Admitted to Australian and New Zealand Intensive Care Units
Statistical Analysis
1 Introduction
This is an explanation of the statistical analysis for the study examining the epidemiology and outcomes of patients admitted to ICU with acute hypoxaemic respiratory failure (AHRF).
2 Methods
This is a retrospective study using data from the Australian and New Zealand Intensive Care Society (ANZICS) adult ICU patient database (APD). This manuscript has been prepared and reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement.
2.1 Patient Population
We included all patients in the ANZICS APD from 01/01/2005 to 31/12/2022. Patients were categorised into PF ratio subgroups according to the severity definitions below.
2.2 Aims and Objectives
- Determine the association between PaO2:FiO2 ratio and hospital mortality.
- Determine the hospital mortality trends over time across PaO2:FiO2 ratio categories.
- Determine the admissions to ICU with AHRF over time across PaO2:FiO2 ratio categories.
- Determine the trends in proportion of hospital outcomes (Hospital mortality, ICU mortality, ICU length of stay, hospital length of stay, discharge home, nursing home/palliation, rehabilitation, other acute care hospital, other hospital ICU) over time by AHRF categories.
- Determine the association between PaO2:FiO2 ratio and hospital mortality in pre-specified subgroups (Level of ventilatory support, gender, age categories, country, hospital type, admission diagnosis, presence of treatment limitations at ICU admission)
- Determine the time to death of each AHRF categories.
- Determine the time to death of time categories.
2.3 Definition of Acute Hypoxaemic Respiratory Failure
Currently, there is heterogeneity in the precise definition of Acute Hypoxaemic Respiratory Failure (AHRF). For this study, we have defined AHRF as a PaO2:FiO2 ratio < 300 mmHg. AHRF can further be categorised as mild (PaO2:FiO2 200-300 mmHg), moderate (PaO2:FiO2 100-200 mmHg) and severe (PaO2:FiO2 <100 mmHg). Further categorsiation can be made into moderate-severe (PaO2:FiO2 100-150 mmHg).
2.4 Statistical Analysis
Baseline ICU- and patient-level characteristics and unadjusted outcomes were summarized using standard descriptive statistics. Continuous variables were reported as either means with standard deviation or medians and interquartile range, and categorical variables reported as number and percentages.
2.4.1 Covariate Selection for Multivariable Regression Models for Adjusted Outcomes
A directed acyclic graph (DAG) was used to generate a minimum adjustment set of covariates for the causal pathway from AHRF to hospital outcomes. Covariates were selected based on potential mechanistic associations with other variables. In addition to this, variables were selected if they increased the precision of the estimate. All back-door paths were closed and a minumum adjustment set of selected variables were used for that particular model. For all outcomes, baseline variables that were included in the model included PaO2:FiO2 ratio, chronic respiratory disease, chronic cardiovascular disease, frailty, smoking intensity and severity of illness scores.
2.4.2 Model Fitting
To analyse the association of AHRF and hospital outcomes, a multivariable, hierarchical logistic regression model was used with patients nested within sites and site treated as a random effect. Predicted probabilities were generated from the model output and displayed graphically as the effect on hospital outcomes by either the continuous PaO2:FiO2 ratio or time. The association between either PaO2:FiO2 ratio or time and hospital outcomes was modeled using restricted cubic splines with 4 knots to allow for non-linear association. Changes over time were described with the estimand of Absolute Risk Reduction (with 95% CI). To determine whether changes over time differed between different AHRF categories, an interaction term between time and the AHRF categories was fitted.
2.4.3 Time-to-Event Analysis
Time to death was analysed using Cox-proportional hazards regression with random effects (frailty model), with covariates included as determined by the minimum adjustment set from the previously constructed DAG. The results are presented as a Kaplan-Meier curve generated from the Cox Proportional Hazards regression with the HR (95% CI) for each AHRF category compared to no AHRF (“None” category). To further investigate the change over time, the study period was divided into 3 cohorts (2005-09, 2010-14, 2015-19, 2020-2022). These were presented as Kaplan-Meier curves from the Cox regression, with the HR (95% CI) comparing each time epoch to the initial time (2005-09). Each AHRF category is displayed as a separate Kaplan-Meier curve to examine the change within that category over time.
As there were >1,500,000 patients in the dataset, a 2-sided p-value of 0.001 was used for statistical significance. Given that there is an increased risk of Type-1 error with multiple testing, the results of the secondary objectives should be viewed as exploratory. Hence, no adjustment for multiplicity was used. Only patients with complete data for all covariates were included in the analysis. Statistical analyses were performed using R Version 4.3.1 (R Core Team, R Foundation for Statistical Computing, Vienna, Austria) and RStudio Version 2023.06.1 (Posit Software, PBC, Boston, MA). Packages used for analysis included tidyverse, data.table, gtsummary, lme4, survival and ggsurvfit.
2.5 Subgroup Analysis
Patients were analysed for the association of PF ratio and in-hospital mortality in the following subgroups: Receiving invasive ventilation on day 1, receiving invasive ventilation at any time during the ICU admission, receiving ECMO during the ICU admission, levels of ventilatory support (none, non-invasive ventilation, invasive ventilation, extra-corporeal membrane oxygenation), gender, age categories, admission diagnosis (medical, surgical, cardiac, neurosurgery/trauma, COVID pneumonitis), and treatment limitation status on ICU admission.
2.6 Sensitivity Analysis
Post-hoc sensitivity analyses were performed on the pre-specified variables of severity of illness scores, and for duration of invasive ventilation. In the first sensitivity analysis, modelling was repeated substituting the APACHE III score for APACHE II, ANZROD and SOFA scores. In the second sensitivity analysis, modelling was repeated on the pre-specified variables of invasive ventilation of more than 12 hours.
3 Results
Of 1,560,221 patients admitted to 211 ICUs during the study period, 826,106 (52.9%) patients had acute hypoaxemic respiratory failure. Of this cohort, 424382 (27.2%) had mild AHRF, 324,630 (20.8%) had moderate AHRF and 77,094 (4.9%) had severe AHRF.
3.1 Patient Demographics
Below are the demographic tables. This table has the 4 AHRF categories (none, mild, moderate, severe) to look at the breakdown of patients within each AHRF category.
| Characteristic | Overall1 | Acute Hypoxaemic Respiratory Failure Category | |||
|---|---|---|---|---|---|
| None (PF >300)1 | Mild (PF 200-300)1 | Moderate (PF 100-200)1 | Severe (PF < 100)1 | ||
| Number of Patients | 1,560,221 | 734,115 | 424,382 | 324,630 | 77,094 |
| Median Age in Years (IQR) | 66 (53-75) | 65 (49-75) | 67 (56-76) | 66 (55-75) | 65 (52-75) |
| Age Category, Years | |||||
| <44 | 241,138 (15%) | 141,470 (19%) | 47,537 (11%) | 39,964 (12%) | 12,167 (16%) |
| >84 | 112,518 (7.2%) | 54,908 (7.5%) | 31,478 (7.4%) | 21,151 (6.5%) | 4,981 (6.5%) |
| 45-64 | 474,408 (30%) | 217,225 (30%) | 130,078 (31%) | 102,607 (32%) | 24,498 (32%) |
| 65-84 | 729,623 (47%) | 319,327 (44%) | 214,624 (51%) | 160,349 (49%) | 35,323 (46%) |
| Gender | |||||
| Female | 641,352 (41%) | 326,027 (44%) | 164,252 (39%) | 122,366 (38%) | 28,707 (37%) |
| Male | 918,260 (59%) | 407,783 (56%) | 259,945 (61%) | 202,173 (62%) | 48,359 (63%) |
| Intersex/Indeterminate | 390 (<0.1%) | 205 (<0.1%) | 112 (<0.1%) | 54 (<0.1%) | 19 (<0.1%) |
| Unknown | 219 (<0.1%) | 100 (<0.1%) | 73 (<0.1%) | 37 (<0.1%) | 9 (<0.1%) |
| Median APACHE II Score (IQR) | 15 (11-21) | 14 (10-18) | 15 (12-20) | 18 (14-24) | 23 (18-31) |
| Median APACHE III Score (IQR) | 52 (39-68) | 47 (35-61) | 52 (40-67) | 60 (46-78) | 76 (58-102) |
| Median ANZROD (IQR) | 0.09 (0.18) | 0.06 (0.13) | 0.08 (0.16) | 0.14 (0.21) | 0.27 (0.29) |
| Median SOFA (IQR) | 4 (2-6) | 3 (1-4) | 4 (3-6) | 5 (4-7) | 7 (5-10) |
| Admission Diagnosis | |||||
| Medical | 478,338 (31%) | 176,685 (25%) | 120,753 (29%) | 136,715 (43%) | 44,185 (59%) |
| Post-Operative | 432,367 (28%) | 243,005 (34%) | 121,506 (29%) | 59,472 (19%) | 8,384 (11%) |
| Sepsis | 113,516 (7.5%) | 47,693 (6.7%) | 30,512 (7.4%) | 27,028 (8.5%) | 8,283 (11%) |
| Trauma/Neurosurgery | 164,707 (11%) | 105,030 (15%) | 38,829 (9.4%) | 18,023 (5.7%) | 2,825 (3.8%) |
| Cardiac Surgery | 330,028 (22%) | 139,524 (20%) | 102,972 (25%) | 76,061 (24%) | 11,471 (15%) |
| COVID Penumonitis (Proven) | 4,451 (0.3%) | 249 (<0.1%) | 600 (0.1%) | 2,340 (0.7%) | 1,262 (1.6%) |
| Admission Source | |||||
| Emergency Department | 379,997 (24%) | 159,769 (22%) | 97,029 (23%) | 95,474 (29%) | 27,725 (36%) |
| Operating Theatre/Recovery | 897,931 (58%) | 474,998 (65%) | 254,818 (60%) | 146,535 (45%) | 21,580 (28%) |
| Ward | 188,779 (12%) | 61,825 (8.4%) | 48,189 (11%) | 58,167 (18%) | 20,598 (27%) |
| ICU, Same Hospital | 2,088 (0.1%) | 769 (0.1%) | 582 (0.1%) | 543 (0.2%) | 194 (0.3%) |
| Other Hospital | 90,715 (5.8%) | 36,435 (5.0%) | 23,587 (5.6%) | 23,743 (7.3%) | 6,950 (9.0%) |
| Direct from Home | 711 (<0.1%) | 319 (<0.1%) | 177 (<0.1%) | 168 (<0.1%) | 47 (<0.1%) |
| Hospital Type | |||||
| Tertiary | 738,248 (47%) | 339,133 (46%) | 200,430 (47%) | 160,733 (50%) | 37,952 (49%) |
| Metropolitan | 243,919 (16%) | 99,911 (14%) | 65,054 (15%) | 60,547 (19%) | 18,407 (24%) |
| Rural / Regional | 145,760 (9.3%) | 58,204 (7.9%) | 39,893 (9.4%) | 37,813 (12%) | 9,850 (13%) |
| Private | 432,294 (28%) | 236,867 (32%) | 119,005 (28%) | 65,537 (20%) | 10,885 (14%) |
| Chronic Respiratory Disease | 113,366 (7.3%) | 32,308 (4.4%) | 36,569 (8.6%) | 36,332 (11%) | 8,157 (11%) |
| Chronic CVS Disease | 141,997 (9.1%) | 57,711 (7.9%) | 41,613 (9.8%) | 34,425 (11%) | 8,248 (11%) |
| Chronic Hepatic Disease | 28,451 (1.8%) | 11,995 (1.6%) | 7,628 (1.8%) | 6,821 (2.1%) | 2,007 (2.6%) |
| Chronic Renal Disease | 51,172 (3.3%) | 22,705 (3.1%) | 14,034 (3.3%) | 11,420 (3.5%) | 3,013 (3.9%) |
| Frailty | |||||
| Fit/Well | 279,102 (59%) | 148,491 (64%) | 73,055 (57%) | 48,012 (54%) | 9,544 (51%) |
| Mild | 141,890 (30%) | 64,823 (28%) | 40,847 (32%) | 29,900 (33%) | 6,320 (34%) |
| Moderate | 34,859 (7.4%) | 14,260 (6.1%) | 10,321 (8.1%) | 8,400 (9.4%) | 1,878 (10%) |
| Severe | 13,361 (2.8%) | 5,151 (2.2%) | 3,976 (3.1%) | 3,413 (3.8%) | 821 (4.4%) |
| Diabetes | |||||
| None | 394,910 (73%) | 202,650 (76%) | 105,305 (71%) | 71,654 (69%) | 15,301 (71%) |
| Type-1 Diabetes | 13,832 (2.6%) | 7,974 (3.0%) | 3,185 (2.2%) | 2,214 (2.1%) | 459 (2.1%) |
| Type-2 Diabetes | 125,514 (23%) | 53,514 (20%) | 37,893 (26%) | 28,423 (28%) | 5,684 (26%) |
| Gestational Diabetes | 1,695 (0.3%) | 1,087 (0.4%) | 349 (0.2%) | 213 (0.2%) | 46 (0.2%) |
| Secondary Diabetes/Other | 3,595 (0.7%) | 1,677 (0.6%) | 1,029 (0.7%) | 738 (0.7%) | 151 (0.7%) |
| 1 n; Median (25%-75%); n (%); Mean (SD) | |||||
3.2 Admission Diagnoses
Below are the admission diagnoses and the associated AHRF categories, ordered by the most common.
| Characteristic | Overall1 | Acute Hypoxaemic Respiratory Failure Category | |||
|---|---|---|---|---|---|
| None (PF >300)1 | Mild (PF 200-300)1 | Moderate (PF 100-200)1 | Severe (PF < 100)1 | ||
| Number of Patients | 1,560,221 | 734,115 | 424,382 | 324,630 | 77,094 |
| Cause of AHRF | |||||
| Cardiac Surgery | 330,028 (22%) | 139,524 (20%) | 102,972 (25%) | 76,061 (24%) | 11,471 (15%) |
| Respiratory Disease | 150,592 (9.9%) | 24,239 (3.4%) | 40,901 (9.9%) | 64,270 (20%) | 21,182 (28%) |
| Cardiovascular Disease | 114,115 (7.5%) | 37,821 (5.3%) | 29,509 (7.1%) | 33,109 (10%) | 13,676 (18%) |
| Sepsis | 113,516 (7.5%) | 47,693 (6.7%) | 30,512 (7.4%) | 27,028 (8.5%) | 8,283 (11%) |
| Gastrointestinal Surgery | 215,730 (14%) | 115,486 (16%) | 62,693 (15%) | 32,861 (10%) | 4,690 (6.2%) |
| Other Non-Operative | 247,942 (16%) | 131,470 (18%) | 59,651 (14%) | 46,174 (15%) | 10,647 (14%) |
| Other Post-Operative | 347,033 (23%) | 215,704 (30%) | 88,334 (21%) | 37,796 (12%) | 5,199 (6.9%) |
| COVID Pneumonitis | 4,451 (0.3%) | 249 (<0.1%) | 600 (0.1%) | 2,340 (0.7%) | 1,262 (1.6%) |
| 1 n; n (%) | |||||
3.3 Admission Diagnoses Over Time
Below are the admission diagnoses trends over time, with categorisation by AHRF severity.
3.4 Admissions to ICU with AHRF Over Time
The proportion of AHRF admissions to ICU was 53.6% (559/1,042) in 2005 and 52.0% (66521/127,934) in 2022. The proportion of patients admitted with mild AHRF to ICU was 27.2% (283/1,042) in 2005 and 27.6% (35,306/127,934) in 2022. The proportion of patients admitted with moderate AHRF to ICU was 20.6% (215/1,042) in 2005 and 19.9% (25,508/127,934) in 2022. The proportion of patients admitted with severe AHRF to ICU was 5.9% (61/1,042) in 2005 and 4.5% (5,707/127,934) in 2022.
3.5 Unadjusted Outcomes
3.5.1 All Outcomes
Below are the unadjusted outcomes for the different AHRF categories. The first table contains the outcomes for patients with AHRF compared to those without it, and the second table has the 4 categories (none, mild, moderate, severe).
| Characteristic | Overall1 | Acute Hypoxaemic Respiratory Failure | |
|---|---|---|---|
| No-AHRF (PF >300)1 | AHRF (PF <300)1 | ||
| Hospital Mortality | 142,432 (9.1%) | 39,250 (5.3%) | 103,182 (12%) |
| ICU Mortality | 93,626 (6.0%) | 21,807 (3.0%) | 71,819 (8.7%) |
| Mean ICU Length of Stay in Days (SD) | 3.4 (5.7) | 2.7 (4.4) | 4.0 (6.5) |
| Median ICU Length of Stay in Days (IQR) | 1.9 (1.0-3.7) | 1.6 (0.9-2.9) | 2.1 (1.0-4.5) |
| Mean Hospital Length of Stay in Days (SD) | 15 (73) | 14 (71) | 16 (76) |
| Median Hospital Length of Stay in Days (IQR) | 9 (5-16) | 8 (5-15) | 10 (6-17) |
| Hospital Discharge Destination | |||
| Home | 1,097,895 (70%) | 549,138 (75%) | 548,757 (66%) |
| Nursing Home/Chronic Care/Palliative Care/Rehabilitation | 142,455 (9.1%) | 68,377 (9.3%) | 74,078 (9.0%) |
| Other Acute Care Hospital | 145,532 (9.3%) | 64,363 (8.8%) | 81,169 (9.8%) |
| Other Hospital- ICU | 20,649 (1.3%) | 7,324 (1.0%) | 13,325 (1.6%) |
| Mental Health | 3,100 (0.2%) | 1,721 (0.2%) | 1,379 (0.2%) |
| Hospital in the Home | 5,609 (0.4%) | 2,793 (0.4%) | 2,816 (0.3%) |
| 1 n (%); Mean (SD); Median (25%-75%) | |||
| Characteristic | Overall1 | Acute Respiratory Failure Category | |||
|---|---|---|---|---|---|
| None (PF >300)1 | Mild (PF 200-300)1 | Moderate (PF 100-200)1 | Severe (PF < 100)1 | ||
| Hospital Mortality | 142,432 (9.1%) | 39,250 (5.3%) | 33,946 (8.0%) | 46,214 (14%) | 23,022 (30%) |
| ICU Mortality | 93,626 (6.0%) | 21,807 (3.0%) | 20,218 (4.8%) | 32,502 (10%) | 19,099 (25%) |
| Mean ICU Length of Stay in Days (SD) | 3.4 (5.7) | 2.7 (4.4) | 3.2 (5.5) | 4.6 (6.9) | 6.1 (8.7) |
| Median ICU Length of Stay in Days (IQR) | 1.9 (1.0-3.7) | 1.6 (0.9-2.9) | 1.9 (1.0-3.6) | 2.7 (1.3-5.1) | 3.3 (1.5-7.2) |
| Mean Hospital Length of Stay in Days (SD) | 15 (73) | 14 (71) | 16 (79) | 17 (70) | 18 (79) |
| Median Hospital Length of Stay in Days (IQR) | 9 (5-16) | 8 (5-15) | 9 (6-16) | 10 (6-18) | 11 (5-20) |
| Hospital Discharge Destination | |||||
| Home | 1,097,895 (70%) | 549,138 (75%) | 302,572 (71%) | 208,950 (64%) | 37,235 (48%) |
| Nursing Home/Chronic Care/Palliative Care/Rehabilitation | 142,455 (9.1%) | 68,377 (9.3%) | 39,703 (9.4%) | 28,174 (8.7%) | 6,201 (8.0%) |
| Other Acute Care Hospital | 145,532 (9.3%) | 64,363 (8.8%) | 40,100 (9.5%) | 33,245 (10%) | 7,824 (10%) |
| Other Hospital- ICU | 20,649 (1.3%) | 7,324 (1.0%) | 5,021 (1.2%) | 5,916 (1.8%) | 2,388 (3.1%) |
| Mental Health | 3,100 (0.2%) | 1,721 (0.2%) | 742 (0.2%) | 516 (0.2%) | 121 (0.2%) |
| Hospital in the Home | 5,609 (0.4%) | 2,793 (0.4%) | 1,585 (0.4%) | 1,043 (0.3%) | 188 (0.2%) |
| 1 n (%); Mean (SD); Median (25%-75%) | |||||
3.5.2 Unadjusted In-Hospital Mortality and Continuous PaO2:FiO2 Ratio
The figures below represent the unadjusted outcomes.
3.5.3 Unadjusted Trends in Hospital Outcomes Over Time by AHRF Categories
3.6 Adjusted Outcomes
3.6.1 Primary Outcome: Conditional Probability of In-Hospital Mortality and AHRF Status
Adjusted mortality
3.6.2 Primary Outcome: Conditional Probability of In-Hospital Mortality and Continuous PaO2:FiO2 Ratio
Adjusted mortality increased as PaO2:FiO2 Ratio decreased (Adjusted ARR -3.598 (95%CI: -3.719 - -3.478)). This means that for every decrease of 100 in the PaO2:FiO2 Ratio, the conditional hospital mortality will increase by 3.598%. However, this effect is non-linear, as seen from the figure below.
3.6.3 Secondary Outcomes: Trends in Conditional Probability of Outcomes Over Time by AHRF Categories
The following details the secondary hospital outcomes both descriptively and graphically.
Overall hospital mortality is decreasing over time. Over the study period, there was a reduction in adjusted hospital mortality from 13.278 (95% CI: 13.092 - 13.464) to 8.2 (95% CI: 8.128 - 8.271). This was an absolute risk reduction (ARR) of 5.079 (95% CI: 5.077 - 5.081). This change in outcome for each AHRF category over time were all significant (all p<0.001 for interactions compared to None category of AHRF), and the comparisons between each category and no-AHRF were all significant (p<0.001).
Overall ICU mortality is decreasing over time. Over the study period, there was a reduction in adjusted ICU mortality from 8.629 (95% CI: 8.472 - 8.786) to 5.538 (95% CI: 5.478 - 5.598). This was an absolute risk reduction (ARR) of 3.091 (95% CI: 3.089 - 3.093). This change in outcome between the categories of AHRF over time were all significant (all p<0.001 for interactions compared to None category of AHRF).
Overall ICU length of stay is decreasing over time. Over the study period, there was a reduction in adjusted ICU LOS from 3.91562 (95% CI: 3.89709 - 3.93416) to 3.26592 (95% CI: 3.25883 - 3.27302). This was an absolute risk reduction (ARR) of 0.6497 (95% CI: 0.6495 - 0.6499). This change in outcome between the categories of AHRF over time were not significant (95% CI crossed 0 for interactions compared to None category of AHRF).
Overall ICU length of stay is decreasing over time. Over the study period, there was a reduction in adjusted ICU LOS from 17.96259 (95% CI: 17.91665 - 18.00853) to 14.53623 (95% CI: 14.51806 - 14.5544). This was an absolute risk reduction (ARR) of 3.42636 (95% CI: 3.42586 - 3.42685). This change in outcome between the categories of AHRF over time were significant (95% CI did not cross 0 for interactions compared to None category of AHRF).
Overall discharge home is increasing over time. Over the study period, there was a reduction in adjusted hospital mortality from 66.744 (95% CI: 66.513 - 66.976) to 70.657 (95% CI: 70.569 - 70.746). This was an absolute risk reduction (ARR) of -3.913 (95% CI: -3.916 - -3.911). This change in outcome between the categories of AHRF over time were all not significant (all p>0.001 for interactions compared to None category of AHRF).
Overall discharge to a facility is increasing over time. Over the study period, there was a reduction in adjusted hospital mortality from 8.973 (95% CI: 8.887 - 9.06) to 9.095 (95% CI: 9.062 - 9.128). This was an absolute risk reduction (ARR) of -0.122 (95% CI: -0.123 - -0.121). This change in outcome between the categories of AHRF over time were all not significant (all p>0.001 for interactions compared to None category of AHRF).
Overall discharge to a different acute care hospital is decreasing over time. Over the study period, there was a reduction in adjusted hospital mortality from 11.47 (95% CI: 11.392 - 11.548) to 8.721 (95% CI: 8.691 - 8.751). This was an absolute risk reduction (ARR) of 2.749 (95% CI: 2.748 - 2.75). This change in outcome between the categories of AHRF over time were all not significant (all p<0.001 for interactions compared to None category of AHRF), with the exception of the severe category of AHRF.
3.6.4 Secondary Outcomes: Time to Death (Adjusted Analysis) Truncated at 1 Year
3.6.4.1 Cox Proportional Hazards
| Characteristic | HR1 | 95% CI1 | p-value |
|---|---|---|---|
| AHRF Category | |||
| None | — | — | |
| Mild | 1.14 | 1.12, 1.15 | <0.001 |
| Moderate | 1.28 | 1.27, 1.30 | <0.001 |
| Severe | 1.48 | 1.46, 1.51 | <0.001 |
| 1 HR = Hazard Ratio, CI = Confidence Interval | |||
| Characteristic | HR1 | 95% CI1 | p-value |
|---|---|---|---|
| Time Category | |||
| 2005-09 | — | — | |
| 2010-14 | 0.98 | 0.97, 1.00 | 0.072 |
| 2015-19 | 1.06 | 1.04, 1.08 | <0.001 |
| 2020-22 | 1.10 | 1.08, 1.12 | <0.001 |
| 1 HR = Hazard Ratio, CI = Confidence Interval | |||
| Characteristic | HR1 | 95% CI1 | p-value |
|---|---|---|---|
| Time Category | |||
| 2005-09 | — | — | |
| 2010-14 | 0.95 | 0.92, 0.98 | 0.004 |
| 2015-19 | 1.0 | 0.96, 1.03 | 0.8 |
| 2020-22 | 0.98 | 0.94, 1.01 | 0.2 |
| 1 HR = Hazard Ratio, CI = Confidence Interval | |||
| Characteristic | HR1 | 95% CI1 | p-value |
|---|---|---|---|
| Time Category | |||
| 2005-09 | — | — | |
| 2010-14 | 0.99 | 0.95, 1.03 | 0.5 |
| 2015-19 | 1.08 | 1.04, 1.12 | <0.001 |
| 2020-22 | 1.12 | 1.07, 1.16 | <0.001 |
| 1 HR = Hazard Ratio, CI = Confidence Interval | |||
| Characteristic | HR1 | 95% CI1 | p-value |
|---|---|---|---|
| Time Category | |||
| 2005-09 | — | — | |
| 2010-14 | 1.00 | 0.97, 1.03 | >0.9 |
| 2015-19 | 1.08 | 1.05, 1.12 | <0.001 |
| 2020-22 | 1.15 | 1.11, 1.19 | <0.001 |
| 1 HR = Hazard Ratio, CI = Confidence Interval | |||
| Characteristic | HR1 | 95% CI1 | p-value |
|---|---|---|---|
| Time Category | |||
| 2005-09 | — | — | |
| 2010-14 | 1.03 | 0.98, 1.07 | 0.2 |
| 2015-19 | 1.21 | 1.16, 1.26 | <0.001 |
| 2020-22 | 1.33 | 1.27, 1.40 | <0.001 |
| 1 HR = Hazard Ratio, CI = Confidence Interval | |||
3.6.4.2 Kaplan-Meier Curves
3.7 Subgroup Analysis (Adjusted Analysis)- Graphical Analysis
3.8 Subgroup Analysis (Adjusted Analysis)- Non-Graphical Analysis
3.9 Sensitivity Analysis
3.9.1 Different Severity of Illness Scores
3.9.2 Using Patients with > 12 Hours Invasive Ventilation
